Drugs and Adverse Drug Reactions: Title and subTitle BreakHow Worried Should We Be?

Physicians can hardly pick up a medical journal or a newspaper today without reading about some new medication, and how it promises to completely change the course of a disease or relieve some troublesome symptom. Indeed, the wonders of pharmacology are numerous. It is clear, for example, that after a myocardial infarction patients will live longer if they take β-blockers ¹ and that patients with congestive heart failure live longer and feel better when they take angiotensin-converting enzyme inhibitors.² However, medications are a double-edged sword.

Much of the recent work on problems with medications has focused primarily on errors in medication use, which are important.³ But, adverse drug reactions (ADRs) that are not preventable given our current state of knowledge are a more common problem, with a greater human burden. In this issue of JAMA, Lazarou and colleagues⁴ attempt to assess the extent of this problem. They performed a meta-analysis of the prospective studies evaluating the incidence of ADRs in hospitalized patients. Even after excluding errors in drug administration, the authors found that ADRs may be the fourth to sixth leading cause of death, and that drug-related injuries occur in 6.7% of hospitalized patients. These data suggest that health care practicioners may miss or pass over many ADRs that occur, even among fatal events.

Can these data be correct? There are a number of concerns about the way the study was done, although the authors adhered to the generally accepted criteria for meta-analyses. ⁵ First, an inherent limitation of meta-analysis is that combining the results of small, heterogeneous studies does not necessarily bring one closer to truth, particularly if the processes used to identify and to validate the presence of the events were heterogeneous. ⁶ Second, the hospitals studied are probably not representative of hospitals at large. Such studies are more likely to be conducted in academic, tertiary care hospitals; these hospitals have sicker patients, and these patients have more ADRs. Another issue is whether the sites of care sampled within the institutions were representative of the institutions. All these factors could inflate the incidence estimate. It is also surprising that the ADR rate remained constant over time, despite increasing patient acuity and use of larger numbers of medications. Some other data⁷ suggest that the frequency of problems with medications actually may be increasing.

Nonetheless, these data are important, and even if the true incidence of ADRs is somewhat lower than that reported by Lazarou et al,⁴ it is still high, and much higher than generally recognized. Why is this the case? One reason is that hospitals have had strong incentives not to identify too many of these events. Reporting large numbers of adverse events and any serious preventable event brings intense scrutiny from regulators and the public. Thus, most hospitals have relied on spontaneous reporting, which only identifies about 1 in 20 adverse reactions and leads to the perception that injuries from ADRs are less common than they really are.⁸ Also, less research has been done in this area compared with other major causes of death, such as heart disease or cancer. No single specialty or organ system is involved, the Food and Drug Administration is not a funding agency, and research funding for this important area has been scarce.

Another issue is whether tracking nonpreventable drug-related injuries is important, especially after it is known that a specific drug can cause a specific reaction. It is, for several reasons: First, avoiding administration of the same medication to the patient in the future requires knowing and documenting that the patient had a previous allergy or sensitivity. When a patient develops an allergy or sensitivity, it is often not recorded, and patients receive drugs to which they have known allergies or sensitivities with disturbing frequency.³ Second, many events not preventable today will likely be preventable in the future, by one of a variety of mechanisms. For instance, design of new and safer drugs is spurred by such data. A good example is development of fexofenidine to replace terfenidine. Terfendine was associated with cardiac arrhythmias including torsades de pointes, particularly when used in combination with several other commonly used medications, including erythromycin.⁹ Substitution of the new agent—a terfenidine metabolite that has the therapeutic properties but not the adverse consequences of terfenidine—seems likely to dramatically reduce this risk. Moreover, it is likely that clinicians will improve their ability to predict which patients will experience adverse consequences from specific drugs.

If hospitals are to monitor for ADRs, what approach should they use? Chart review is too expensive to be practical on a routine basis.³ Fortunately, computer surveillance can be used to assist in finding adverse drug events, and this approach is much more efficient than chart review. ^{10 - 11} Today, most hospitals could not immediately implement such a system given the present state of their information systems, but they should be able to do so soon.

But why should hospitals invest in comprehensive monitoring for ADRs, given today's multiple competing priorities? One reason is new regulations, which are being developed by the Health Care Financing Administration.¹² These regulations, released with a request for comment in the Federal Register¹² in November, would require hospitals to routinely monitor for adverse drug events and would impose sanctions if they fail to do so. However, as currently written, the regulations seem untenable for a number of reasons, including the requirement for hospitals to perform expensive routine chart review. If rewritten, these regulations could provide a major impetus for quality. For example, hospitals could be required to demonstrate that they routinely measure adverse drug events and medication error rates.

For all medications, a key issue is whether the benefits outweigh the risks. The answer, as demonstrated by large numbers of randomized controlled trials, is yes, but that there must be more attention given to the risk side of the equation. Only after drugs leave the trial setting and are used in sicker patients do their true risks become apparent. Although some risks are inevitable, they can be significantly reduced, and learning more about these risks will make this possible.